Compelling Limit of Detection Examples

Question

I am looking for guidance in talking about this issue in customer demos.  I know we have Metacrate in the training starter but I am looking for something a bit more intuitive.

Does anyone have a good demo for this that uses something other than Metacrate?

martindemel · Answer

Hi Wendy,

that's what I used in the past to talk about LOD/LOQ.

I added to the Metacrate Example some additional hidden columns to show how it can be treated in different ways and to compare results. Happy to answer questions. I added walk throughs in the journal.

JMPProDay_LimitOfDetection.zip

Phil_Kay · Answer

Hi Wendy,

I don't have any other examples. I would be interested to know what you don't like about the Metacrate example. I suspect it would be possible to find other DOE examples in the literature but I think they would be more or less the same. Or you could probably simplify the Metacrate demo by just showing the end result of the different methods first (Great Demo) and check to see if the audience is interested.

I have to say that I have not experienced anyone buying JMP Pro for this capability. It is not the most exciting feature. I think some existing Pro users are now using this.

However, there is a lot of this data around, especially in chemical and pharma. And scientists are often very uncomfortable about the existing methods for dealing with it. If they are feeling that pain, then the capability should be quite compelling. This is a fairly clear case where we can say that this method is the "correct" method and anything else that they have been doing is "wrong." (Although I would guide them to reaching that conclusion for themselves rather than stating it explicitly)

Do you have a specific customer in mind? Or are you hoping to generate some interest by showing this to a number of customers and prospects? Personally, I don't think a "fishing expedition" based on this feature would be too productive. I think it would be worth asking questions when talking to existing customers to see if they recognise problem.

* Do you ever need to model data where some values are unknown because they are below LOQ or LOD?
* Are you comfortable with your methods for modelling this kind of data?
* What kind of decisions might you make based on this modelling?
* How important are these decisions?
* ...